The Games That Designed Themselves

It is almost impossible to imagine Ape Out without an ape, but that is exactly how it started. Designer Gabe Cuzzillo was first making a time-looping stealth game where the player used push and grab mechanics to move along walls.

Once guards entered the design, those same mechanics naturally worked on them. The most interesting part became grabbing guards, holding them hostage, and throwing them into walls. So the project changed direction. The stealth and time loops fell away, and the bald human character became a rampaging 300-pound gorilla.

That is the design method often called "follow the fun". The idea sounds simple: ignore the plan when the game itself starts pointing somewhere better. Play the prototype, notice what is actually interesting, and be willing to rebuild around that discovery.

Into the Breach started as a more standard Advance Wars-style tactics game. Enemies chose attacks at random and hid their intentions until their turn. Then one enemy showed exactly where it would attack by highlighting the target tile.

That single idea became the most enjoyable part of the prototype. Subset Games focused the rest of the design around telegraphed enemy attacks, and that decision shaped nearly everything else. If the player knows where attacks are going, moving units out of danger is too easy. So the game becomes about protecting static buildings. Then it becomes about pushing enemies away from those buildings. Then it becomes about tricking enemies into hitting each other.

This is why designers sometimes say that a game, in some sense, designs itself. The studio still makes every decision, but the best decisions come from listening to what the prototype has already made exciting.

The idea does not appear from nowhere. It usually begins as something loose, ordinary, or even borrowed, and then changes once it becomes playable.

Crypt of the NecroDancer began when Ryan Clark wondered what would happen if Spelunky's quick decision-making were placed inside a turn-based dungeon crawler. The prototype gave the player only a second to choose each move. Once he played it, the rhythm was obvious, and the dungeon crawler became a music-driven roguelike.

Rocket League found one of its defining mechanics through a similarly practical accident. While building its predecessor, Psyonix wanted a speed boost for battling cars and applied a physics force to the back of the vehicle. During testing, players discovered that the same force could launch cars through the air, adding a new dimension of depth. The studio kept it.

There is a long history of accidents becoming features. Hideki Kamiya found a bug in Onimusha: Warlords that let players juggle enemies in the air by slashing them repeatedly. The bug was fixed there, but Kamiya developed the idea further and made aerial juggling central to Devil May Cry.

The important step is building something that can answer back. During the process of coding, playing, and breaking the first version, new ideas appear. The designer has to notice them, explore them, and accept that they may not match the original pitch.

Gunpoint is a clear example. It was once about a robot in space who dropped fridges on people. Then it became a side-scrolling, Deus Ex-inspired stealth game with a possible hacking minigame about rewiring buildings. As soon as Tom Francis prototyped the rewiring mechanic, the actual game began to emerge. The mechanic wanted to be a puzzle game, so the whole project turned around it.

This kind of change tends to happen most dramatically near the beginning, when the game is still malleable. But the method does not disappear once the project starts to settle into a clearer shape.

Once a game has a core, following the fun can still guide content, feedback, and direction. Jonathan Blow has described Braid's puzzles as showcases for the unexpected consequences of a time-travel engine. The designer becomes a curator: finding the interesting results, cleaning them up, and presenting them so players can enjoy them.

SpyParty changed through player behavior. Chris Hecker watched players discover exploits and unintended ways to play. Instead of simply removing those rough edges, he leaned into them, making mind games and psychological tricks a more official part of the experience.

Subnautica also found part of its identity through development. The team did not begin with the intention of making a horror game, but as the project grew, the ocean, isolation, darkness, and unknown creatures kept pushing the experience in that direction. The game told the team where it wanted to go.

This process can make production hard to predict. That is one reason it is often easier to find in independent development than in tightly scheduled blockbuster production.

Tom Francis ran into this on Heat Signature. After Gunpoint, he hoped the fuzzy idea of going inside spaceships would naturally lead to a great design. It did eventually, but only after years of building systems: ship generation, AI, combat, a galaxy map, an economy, and more. It took a long time to discover that on-ship combat was the most interesting part.

That is why the fuller phrase matters: fail fast, then follow the fun. Marc LeBlanc used that framing to describe the value of throwing something together quickly, seeing what works, and letting even a failed attempt point toward the next one. Failure is useful when it happens before too much time has been spent protecting the wrong idea.

One way to make the method practical is to shorten the distance between idea and play. Game jams are built around that pressure. A weekend prototype can test whether an idea has life without asking the designer to commit months or years to it. Baba is You began that way, from a fast event where the idea could be tried, judged, and discarded if it did not work.

Rapid prototyping tools help for the same reason. So do paper prototypes, LEGO, Dreams, or any medium that lets a designer test the rule before polishing the surface. If most of the game already exists and the challenge is content, custom level creation tools can speed up discovery and let more people contribute ideas.

Placeholder art, music, and story can also keep the team focused on mechanics. The earliest game jam prototype of Don't Starve represented its hero with Link from Zelda. The point was not the character art. The point was to test whether the survival loop was worth following.

Total freedom can make the search too wide. It can help to keep one part of the game fixed while everything else stays open to change.

Journey used an iterative process, but thatgamecompany held onto one stable idea: the game would explore the theme of love. That constant gave the team a direction to follow and narrowed the set of discoveries that were worth developing.

A fixed theme, feeling, constraint, or player fantasy can act like a compass. The game can still surprise its creators, but the team has a way to decide which surprises belong.

The main lesson is to stop waiting for the perfect game idea to arrive fully formed. It is easy to look at Ape Out, Crypt of the NecroDancer, Crashlands, or any other sharply focused game and assume it began as a clean flash of insight.

In reality, these games came from designers who started building. They tried ideas, played prototypes, made mistakes, found bugs, noticed the interesting parts, and followed those paths. Their skill was not only inventing the first idea. It was knowing how to listen to the game, how to fail quickly, how to follow the strongest thread, and how to pull scattered discoveries into something coherent.

So do not wait for the perfect idea. Build something small enough to play. Listen to what it is telling you. Follow the fun, and the game may start designing itself.